[Apache TVM Discuss] [Questions] Can't build module after applying pass MakePackedAPI

[Sen Yang via Apache TVM Discuss <no...@discuss.tvm.ai>]

Hello, I am studying tir and tir pass recently. And I found after I applying pass `MakePackedAPI`, the function `tvm.build` will throw an error:

```
TVMError:
---------------------------------------------------------------
An error occurred during the execution of TVM.
For more information, please see: https://tvm.apache.org/docs/errors.html
---------------------------------------------------------------
  Check failed: (it != info_map_.end()) is false: Load/Store of buffer arg_type_ids (0x55d15e9be390) occurred before its declaration.
```

I am curious about the reason why I can't do that?

Here is my code:

```python
import tvm
from tvm import te
def check_packed_func(target="llvm"):
    ib = tvm.tir.ir_builder.create()

    m = n = k = 16

    a = te.placeholder((m, k), name="a", dtype="float64")
    b = te.placeholder((k, n), name="b", dtype="float64")
    k = te.reduce_axis((0, k), name="k")
    c = te.compute((m, n), lambda i, j: te.sum(a[i, k] * b[k, j], axis=k), name="c")

    a_buffer = tvm.tir.decl_buffer(
        a.shape, a.dtype, name="a_buffer", offset_factor=1, strides=[te.var("s1"), 1]
    )
    b_buffer = tvm.tir.decl_buffer(
        b.shape, b.dtype, name="b_buffer", offset_factor=1, strides=[te.var("s2"), 1]
    )
    c_buffer = tvm.tir.decl_buffer(
        c.shape, c.dtype, name="c_buffer", offset_factor=1, strides=[te.var("s3"), 1]
    )

    with ib.for_range(0, 10, "i", kind="parallel"):
        ib.emit(tvm.tir.call_packed("tvm.test_matmul", a_buffer, b_buffer, c_buffer))

    stmt = ib.get()

    # Construct a valid IRModule to be lowered:
    mod = tvm.IRModule.from_expr(tvm.tir.PrimFunc([a_buffer, b_buffer, c_buffer], stmt))

    target = tvm.target.Target(target)
    mod = tvm.tir.transform.Apply(lambda f: f.with_attr("target", target))(mod)
    mod = tvm.tir.transform.Apply(lambda f: f.with_attr("global_symbol", "main"))(mod)
    mod = tvm.tir.transform.MakePackedAPI()(mod)

    return mod

mod = check_packed_func()
tvm.build(mod)
```





---
[Visit Topic](https://discuss.tvm.apache.org/t/cant-build-module-after-applying-pass-makepackedapi/10814/1) to respond.

You are receiving this because you enabled mailing list mode.

To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/9b29fad834e9179b143b0327c7e5ae47d0093b39885e151382cc33dbb8eee8cd).